Journal of Chromatography B, 791 (2003) 103–109www.elsevier.com/ locate/chromb

S imultaneous determination of gemcitabine and its metabolite inhuman plasma by high-performance liquid chromatography

a a , b*¨ ˘Bilal Yılmaz , Yucel (Yasar) Kadıoglu , Yılmaz Aksoya ¨Ataturk University, Faculty of Pharmacy, Department of Analytical Chemistry, 25240,Erzurum, Turkey

b ¨Ataturk University, Faculty of Medicine, Department of Urology 25240,Erzurum, Turkey

Received 21 November 2002; received in revised form 21 January 2003; accepted 4 March 2003

Abstract

Gemcitabine (dFdC) is a pyrimidine antimetabolite with broad spectrum activity against tumors. In this paper, anormal-phase high-performance liquid chromatographic method was developed for the determination of the parent drug(dFdC) and its metabolite (dFdU) in human plasma. The described sample preparation procedure for determination of dFdCand dFdU is rapid, sensitive, reproducible and simple. The linear regression equations obtained by least square regression

21 24 21method, were area under the curve50.0371 concentration (ng ml )1192.53 and 1.05?10 concentration (ng ml )21.2693 for dFdC and dFdU, respectively. The assay for dFdC and dFdU described in the present report has been applied toplasma samples from a bladder cancer patient. 2003 Elsevier Science B.V. All rights reserved.

Keywords: Gemcitabine; 29-Difluoro-29,29-deoxyuridine

1 . Introduction primary metabolite observed in human plasma andurine [4].

29,2-Difluorodeoxycytidine (gemcitabine, dFdC, Several methods have been reported for the de-Gemzar) is a deoxycytidine analoque [1] with clini- termination of dFdC and dFdU [3–10]. An ELISAcal activity against several solid tumors, including assay [4] has been used for the determination ofovarian cancer, non-small cell lung cancer, head and dFdC and dFdU in human plasma and urine, while

19neck squamous cell carcinoma, pancreatic cancer and F-NMR [5] and HPLC techniques [3,6–10] havebladder cancer [2]. dFdC is a prodrug that undergoes been used for the quantitation of dFdC and dFdU inmetabolism by cytidine deaminase to from an inac- biological fluids. But no HPLC method with a diodetive metabolite (Fig. 1a and b) [3]. The metabolite array detection (DAD) system in bladder cancer has(dFdU, 29-difluoro-29,29-deoxyuridine) was also the been reported in the literature. We wanted to develop

a new HPLC method for the determination of dFdCand its metabolite in plasma.

The aim of this work was to improve a method*Corresponding author. Tel.:190-442-231-1539; fax:190-using a smaller sample volume and a more sensitive442-236-0962.

˘E-mail address: ykadiogli@yahoo.com(Y.(. Kadıoglu). and specific DAD system, and to validate the whole

1570-0232/03/$ – see front matter 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S1570-0232(03)00211-3

104 B. Yılmaz et al. / J. Chromatogr. B 791 (2003) 103–109

2 .3. Chromatographic conditions

The analytical column was Nucleosil 5 NH (52

mm, 2503 4.0 mm). The column temperature was30 8C. CHROMQUEST software was used for on-linedata acquisition. The mobile phase consisted of me-thanol–cyclohexane–1,2-dichloroethane (30:50:20,v /v /v). Themobile phase was filtered through nylonmembrane of 0.45-mm pore size (47 mm filtermembrane, USA). The flow-rate was set at 1.0 ml

21min . The peaks of interest eluted within 10 min.An injection volume of 10ml was used.

2 .4. Preparation of plasma standards and controls

A standard stock solution containing dFdC andFig. 1. Chemical structures of (a) gemcitabine (dFdC) and (b) its dFdU was prepared in methanol at a concentration ofmetabolite (dFdU). 21approximately 50mg ml of each compound. The

stock solutions were stable for at least 3 weeks whenstored refrigerated.

Standard calibration solutions were prepared byanalytical method, according to international guide-

spiking drug-free human plasma with stock standardlines in order to obtain an efficient tool for further

solutions, which were then further diluted to achievepharmacokinetic studies. 21final concentrations of between 0.2 and 50mg ml

21of dFdC and 0.2 and 40mg ml of dFdU. dFdCplasma control samples were prepared from a sepa-

2 . Experimental rate stock solution at concentrations of approximate-21ly 3, 10, 20 and 50mg ml while those of dFdU

212 .1. Instruments were 0.5, 5.0, 20 and 40mg ml .

HPLC was carried out using a Thermoseparations2 .5. Plasma sample preparation procedure

Spectra Series P 4000 gradient pump coupled with aSpectra System UV 6000 LP photodiode array

Individual 0.2-ml aliquots of plasma standards,detection system and a Thermoseparations AS 3000

controls and subject samples were pipetted into 12-autosampler. The detector was set to scan from 200

ml disposable glass tubes. Standard solutions (0.2to 500 nm and had a discrete channel set at 272 nm,

ml) were added, into the plasmas and the solutionswhich was the wavelength used for quantification.

were briefly vortexed. Then 1 ml of isopropanol wasadded, followed by vortex-mixing. The samples were

2 .2. Reagent and standards allowed to sit for 5 min. A 2.5-ml aliquot of ethylacetate was added and vortex-mixed and then the

Gemcitabine and dFdU were kindly provided by samples were centrifuged for 10 min at approximate-Lilly Research (Eli Lilly, Indianapolis, IN, USA). ly 2500g at 48C. The organic phase was transferredHPLC grade methanol, cyclohexane and 1,2-di- to a 5-ml tube and evaporated to dryness at 408Cchloroethane were purchased from Merck. All other under a stream of nitrogen. The dried residues werechemicals were obtained from commercial sources reconstituted in 1 ml of methanol. All samples wereand were of analytical grade. filtered through a Phenomenex membrane of 0.45-

B. Yılmaz et al. / J. Chromatogr. B 791 (2003) 103–109 105

mm pore size before injection. Injection volumes of [11]. Peak purities for dFdC and dFdU were furthersamples and standards (10ml) were performed with confirmed by means of a photo-DAD system.an automatic sample injector. A standard stock solution containing dFdC and

dFdU was prepared in methanol at a concentration of21approximately 50mg ml of each compound. The

stock solutions were stable for at least 3 weeks when3 . Results and discussion stored refrigerated. dFdC concentrations in the work-

ing standard solutions chosen for the calibration3 .1. HPLC measurements and validation curve were 0.2, 2.0, 5.0, 10, 15, 25, 35 and 50mg

21ml while that of dFdU were 0.2, 0.3, 0.5, 1.0, 2.5,21The criteria established for the development of our 5, 10, 20, 30 and 40mg ml . These working

analytical procedure include: (1) using the smallest solutions were made by further dilution of the stockamount of mobile phase possible; (2) restrictingk9 solutions in methanol. dFdC plasma control samplesvalues to between 1 and 10; (3) ionization suppres- were prepared from a separate stock solution atsion between drug molecule and residual amino concentrations of approximately 3, 10, 20 and 50mg

21groups of the surface of silica. ml while those of dFdU were 0.5, 5.0, 20 and21It is known that HPLC–DAD is a highly effective 40mg ml .

screening method. Criterion for identification of the Under the described chromatographic conditionsanalyte is that the maximum absorption wavelength dFdC and dFdU are well separated. Typical chro-in the UV spectrum of the analyte should be the same matograms of blank plasma sample, spiked plasma

21as that of the standard material within62 nm. The sample containing 10mg ml dFdU and dFdC areuse of the photo-DAD also confers the advantage of shown in Fig. 2a–c. The mean retention times wereidentifying the analyte both by retention time and 7.5 and 4.3 min for dFdC and dFdU, respectively. A

21UV spectrum. chromatogram of 10mg ml of dFdC and itsThe following parameters were determined for the metabolite (dFdU) spiked in plasma is shown in Fig.

21validation of analytical method developed for dFdC 3. A chromatogram of 15mg ml of the injectableand dFdU in human plasma; linearity, precision, dosage form of commercial Gemzar spiked in plasmaaccuracy, LOQ, recovery, ruggedness and stability is shown in Fig. 4.

21Fig. 2. Representative chromatograms of (a) blank human plasma without compounds, (b) human plasma standard spiked with 10mg ml21of dFdU, (c) human plasma standard spiked with 10mg ml of dFdC.

106 B. Yılmaz et al. / J. Chromatogr. B 791 (2003) 103–109

21 21Fig. 3. Chromatogram of spiked human plasma sample containing 10mg ml dFdC and its metabolite 10mg ml dFdU.

21Fig. 4. Chromatogram of 15mg ml of the injectable dosage form of commercial Gemzar spiked in human plasma.

3 .2. Linearity The precision and accuracy for each procedurewas determined by spiking dFdC and dFdU into

The linearity of the response for the plasma assay drug-free human plasma at four concentrations. Sixwas established over the concentration range 0.2–50 replicates from each pool were assayed on each of 3

21 21mg ml for each dFdC and 0.2–40mg ml for days so that both intra- and inter-day precision andeach dFdU. Typical correlation coefficients were accuracy could be determined. The results for dFdC.0.99 (Table 1). and dFdU in human plasma are shown in Tables 2

and 3, respectively. For all the concentrations3 .3. Precision and accuracy studied, the RSD values and the relative errors for

the plasma method were,10% and for all con-The precision of a quantitative method is the centrations of compounds the accuracy was.90%.

degree of agreement among individual test resultswhen the procedure is applied repeatedly to multiple 3 .4. Limits of quantitationsamplings. It is measured by repeatedly injecting aready-made sample pool and expressed as relative The LOQ, defined in the presented experiment asstandard deviation of the results. the lowest plasma concentration in the calibration

The precision of the test was evaluated by de- curve that can be measured routinely with acceptabletermining the inter- and intra-day RSD of the precision (RSD,20%)measured peak areas for different concentrations. The lower LOQ for dFdC and dFdU assays were

Table 1Features of the calibration curves of dFdC and its metabolite (dFdU)

Method Sample Plot Wavelength Linear range Regression equation r-1(nm) (mg ml )

HPLC dFdC 8 272.0 0.2–50 y 50.0371x 1 192.532 0.99224dFdU 10 272.0 0.2–40 y 51.05?10 x 2 1.2693 0.996

B. Yılmaz et al. / J. Chromatogr. B 791 (2003) 103–109 107

Table 2Summary of assay precision and accuracy data for dFdC in human plasma

Added Intra-day Inter-day21(mg ml )

Found Precision Accuracy Found Precision Accuracy21 21(mg ml ) RSD (%) relative error (%) (mg ml ) RSD (%) relative error (%)

3.0 2.7 10.38 29.67 2.8 12.40 25.6010 9.7 4.72 22.70 9.8 6.17 22.0020 17.8 5.16 210.90 18.7 4.85 26.2750 47.6 3.52 24.80 47.8 3.72 24.80

Table 3Summary of assay precision and accuracy data for dFdU in human plasma

Added Intra-day Inter-day21(mg ml )

Found Precision Accuracy Found Precision Accuracy21 21(mg ml ) RSD (%) Relative error (%) (mg ml ) RSD (%) Relative error (%)

0.5 0.46 6.37 28.0 0.47 5.37 26.05.0 4.78 4.92 24.4 4.58 13.0 28.4

20 19.66 4.50 23.7 19.12 6.10 24.440 38.80 5.07 23.0 38.48 4.91 23.8

Accuracy5[(found2added) /added]3100; RSD, relative standard deviation.

3 .6. Specificity210.15 and 0.18mg ml , respectively. The limits of

detection for dFdC and dFdU were 0.10 and 0.12mg Specificity is the ability of a method to determine21ml , respectively, at a signal-to-noise ratio of 3. accurately and specifically the analyte of interest in

the presence of other components in a sample matrixunder the stated conditions of the test [13].

3 .5. Recovery Chromatograms representing the separation of theanalytes from the matrix are shown in Fig. 3. No

The absolute recovery was calculated by compar- plasma components were detected at the retentioning peak areas obtained from freshly prepared sam- times for dFdC or dFdU in blank plasma samples.ple extracts with those found by direct injection of dFdC and dFdU were well retained from the voidaqueous standard solutions of the same concentration withk9 values of 2.3 and 0.9, respectively.[12]. The recoveries of dFdC and dFdU from plasma System specificity was gauged by selectivity (a)averaged 91.52 and 94.08%, respectively (Tables 4 and resolution (R ) in sample chromatograms. dFdCs

and 5).

Table 4 Table 5Summary of assay recovery data for dFdC in human plasma Summary of assay recovery data for dFdU in human plasma(n56) (n56)

Concentration Recovery RSD Concentration Recovery RSD21 a 21 a(mg ml ) ratio (%) (%) (mg ml ) rate (%) (%)

3 90.669.40 10.38 0.5 92.965.92 6.3710 97.364.59 4.72 5.0 91.664.51 4.9220 89.164.59 5.16 20 95.664.30 4.5050 95.263.35 3.52 40 96.264.88 5.07

RSD5[(standard deviation/mean)]3100. RSD5[(standard deviation/mean)]3100.a aMean values. Mean values.

108 B. Yılmaz et al. / J. Chromatogr. B 791 (2003) 103–109

Table 6 weeks followed by a rest week. Blood samples wereSystem suitability parameters for dFdC and dFdU in human collected at different times after drug administrationplasma

and blood samples were analyzed immediately. AParameter Metabolite Parent drug typical chromatogram of a plasma sample collected

(dFdU) (dFdC) from the bladder cancer patient 15, 40, 7200 min22Retention time,t (min) 4.3 7.5R after intravenous administration of 1000 mg m

9Capacity factor,k 0.9 2.3 Gemzar is shown in Fig. 5a –d. Fig. 6a and bResolution,R – 8.0S represent the concentration–time profiles of dFdCSelectivity,a – 2.6

and dFdU metabolite in the patient with bladderNumber of theoretical plates,N 3328 3643cancer.

and dFdU were sufficiently well resolved (R . 8.0)s

from each other Table 6.4 . Conclusion

3 .7. Ruggedness and stabilityIn the developed method, 0.2 ml of plasma was

The formal ruggedness test was conducted when used in the sample preparation. Only 10ml of thethe method were validated on another HPLC system reconstituted solution (1 ml) was injected into theby another analyst. Using the optimized parameters, system. This method was utilized in the analysis ofthe method was found to be equally robust. dFdC plasma samples collected from a bladder cancerand dFdU were found to be exceedingly stable patient who was administered Gemzar (1000 mg

22compounds under all the conditions examined. m ). Differences in the pharmacokinetic propertiesof the parent drug (dFdC) and its metabolite (dFdU)

3 .8. Application of the method were observed.The whole HPLC procedure described here for the

The developed method was applied to plasma simultaneous determination of dFdC and dFdU issamples from a bladder cancer patient. A patient widely available in biochemical laboratories.

22with advanced cancer was treated with 1000 mg m In particular, the method has satisfactory spe-Gemzar infused over a 30-min period. The patient cificity, linearity, recovery, accuracy and precisionreceived a Gemzar infusion once per week for 3 range over the concentration range examined.

22Fig. 5. Chromatograms from patient administered a 30-min infusion of 1000 mg m of gemzar (a) patient sample prior to infusion,(b) patient sample at 15 min postinfusion, (c) patient sample at 40 min postinfusion, (d) patient sample at 7200 min postinfusion.

B. Yılmaz et al. / J. Chromatogr. B 791 (2003) 103–109 109

Fig. 6. Profile of the concentrations in patient plasma of (a) dFdC, (b) dFdU determined by the HPLC method.

[6] P.J. Cozzi, D.F. Bajorin, W. Tong, H. Nguyen, J. Scott,A cknowledgementsW.D.W. Heston, G. Dalbagni, Clin. Cancer Res. 5 (1999)2629.

The authors would like to thank the Directorship [7] J.Z. Kerr, S.L. Berg, R. Dauser, J. Nuchtern, M.J. Egorin, L.¨of the University of Ataturk, Erzurum, Turkey, and McGuffey, A. Aleksic, S. Blaney, Cancer Chemother. Phar-

Lilly Research Laboratories (USA) for samples of macol. 47 (2001) 411.[8] C.J.A. Van Moorsel, H.M. Pinedo, K. Smid, E.M. Comijn,gemcitabine (dFdC) and its metabolite (dFdU).

D.A. Voorn, G. Veerman, B. Lakerveld, W.J.F. Van der Vijgh,G. Giaccone, P.E. Postmus, G.J. Peters, Eur. J. Cancer 36(2000) 2420.

R eferences [9] R. Grunewald, H. Kantarjian, M. Du, K. Faucher, P. Taras-soff, W. Plunkett, J. Clin. Oncol. 10 (1992) 406.

[10] G.J. Peters, B. Tonis, M. Clavel, J. Guastella, R. Quick, D.[1] L.W. Hertel, J.S. Kroin, J.W. Misner, J.M. Justin, J. Org.Osborne, C. Mc Daniel, J. Van Maanen, J. Steenbergen,Chem. 53 (1998) 2406.W.J.F. Van der Vijgh, H.M. Pinedo, J. Vermorken, Proc. Am.[2] C.J.A. Van Moorsel, G.J. Peters, H.M. Pinedo, Oncologist 2Assoc. Cancer Res. 31 (1990) 180.(1997) 127.

[11] FDA guideline for submitting samples and analytical data for[3] K.B. Freeman, S. Anliker, M. Hamilton, D. Osborne, P.H.method validations, February 1987.Dhahir, R. Nelson, S.R.B. Allerheiligen, J. Chromatogr. B

[12] P. Hubert, P. Chiap, M. Moors, B. Bourguignon, D.L.665 (1995) 171.Massart, J. Chromatogr. A 665 (1994) 87.[4] J.L. Abbruzzese, R. Grunewald, E.A. Weeks, D. Gravel, T.

´[13] C. Jimenez, R. Ventura, J. Segura, J. Chromatogr. B 767Adams, B. Nowak, S. Mineishi, P. Tarassoff, W. Satterlee,(2002) 341.M.N. Raber, W. Plunkett, J. Clin. Oncol. 9 (1991) 491.

[5] H.T. Edzes, G.J. Peters, P. Noordhuis, Anal. Biochem. 214(1993) 25.